A comparative study of the development of two chemical technologies : polypropylene and epdm rubber

WORKING

PAPER

ALFRED

P.

SLOAN

SCHOOL

OF

MANAGEMENT

A

COMPARATIVE

STUDY

OF

THE

DEVELOPMENT

OF

TWO

CHEMICAL

TECHNOLOGIES:

POLYPROPYLENE

AND

EPDM

RUBBER

by

Michael A.

Rappa

and Koenraad Debackere

May

1990

WP#

3167-90-BPS

^tS

^EIJUL

26

1990 |

MASSACHUSETTS

INSTITUTE

OF

TECHNOLOGY

50

MEMORIAL

DRIVE

CAMBRIDGE,

MASSACHUSETTS

02139

A COMPARATIVE

STUDY

OF

THE

DEVELOPMENT

OF

TWO

CHEMICAL

TECHNOLOGIES:

POLYPROPYLENE

AND

EPDM

RUBBER

by

Michael A.

Rappa

and Koenraad Debackere

May

1990

WP#

3167-90-BPS

Michael

Rappa

isassistantprofessorof

management

in the

Management

ofTechnology Group, atthe Alfred P.Sloan School ofManagement, Massachusetts InstituteofTechnology. Koenraad Debackereis

a doctoral candidate in theVlerick School of Management, Gent University.

The

authors greatly appreciate the adviceand supportof

Han

Van

GoolandJefVincent and theircolleagues at

DSM

Research incarrying outthis work.

An

earlierversion ofthispaper

was

presentedat the

TIMS/ORSA

ABSTRACT

Thispaper examinesthe

dynamic

characteristicsof

R&D

communities and

their role in technological development.

Using

a comparative study of

two

chemical

technologies,

polypropylene

and

epdm

rubber technology, it

empiricallyinvestigates thedevelopmentofthe

two

R&D

communities usingthe

literature

and

patents published

by

polypropylene

and

epdm

researchers as a

sourceofinformation.

The

data allow us to study theinfluence ofdie structural

and

behavioral characteristics of

R&D

communities on

the rate of technical progress, since both technologies have experienced arather differential path of progress.

The

two

chemical technologies also enable us to see

how

the

dynamics

of

R&D

communities

may

change

when

the creations of the

researchers

become

incorporated within a corporate

environment

such as the

chemical industry.

Our

findings point to the

emergence

of a hybrid system, in

which

both managerial control

and

community autonomy

cooperate in the

choice of the directions technological

development

takes.

Moreover,

a

comparison between

polypropylene

and

epdm

sheds further light

on

the

influenceparticular

community

variablespossibly have

on

the rateoftechnical

1

Introduction

The

existence ofinvisible colleges,

embedded

within broaderscientificcommunities, has for

long been a

major

interest ofstudentsof science (Hagstrom. 1965; Griffith andMullins, 1972;

Mullins, 1972;

Ziman,

1984; Price, 1986).

The

notion

"R&D

community"

developed in this paper

seeks toextendthe conceptofan invisiblecollege, tooffera

more encompassing

view, andin

doing so, toexplore itsrelevancetotechnologicaldevelopment.

An

R&D

community

isdefinedasagroup ofindividuals,

composed

ofscientists and

engineers,

who

are

committed

to solving asetofinterrelated scientific and technological problems,

who may

be organizationally andgeographicallydispersed,

and

who

communicate

in

some

way

with each other.

The

ultimate goal for

some

members

ofthe

community

may

be tocreate

new

knowledge, while forothersit

may

be toapply existing

knowledge

inthe creationof

new

products

or processes. Furthermore, the

community

can include individuals

employed

in anytype of

organization, such as universities, private firms,

new

ventures, quasi-public corporations,

and

government

researchinstituteswhereverthey

may

belocated throughout theworld.

Obviously, this conceptofan

R&D

community

extends

beyond

traditionaldisciplinary and

industry boundaries. Itis broaderin nature than the discipline-based scientificcommunities and

thanthe smaller subsets within thosecommunitiesthatare

formed

by theexistenceofthe invisible

college. Rather,the

R&D

community

has aninterdisciplinarycharacter

and

can include researchers

from

a

wide

varietyofacademic specialties. Moreover, withinthe private sector, it includes firms

irrespectiveoftheirstandardindustrial classification.

The

only requirementistheir focus on

some

partoftherelevant

problem

set.Thus, the

community

is definedby the natureofthe problemset

andnot necessarily

by

the

end

product, asis normally thecase with anindustrydefinition.

technology. It is believedthat this approach broadensthe scopeof current thinking

on

technologicaldevelopment

which

tends toemphasize the instrumentalrole ofthe firmin bringing

new

technology tothemarketplace.

Much

academicresearch onthesubjectof technological

development

hasbeendirectedatunderstanding the functioning of firms (and industrialresearch

laboratories in particular), attemptingtounravel therelationships between

numerous

organizational

variables

and

successin developing

new

technologies (Allen, 1977; Roberts, 1984). There is

no

doubt thatprevious

work

along this lineofinquiry has yielded importantinsights aboutthe

management

oftechnologicaldevelopment.

However,

infocusing

on

the firm in isolationofthe

broader environmentin

which

technologicaldevelopmentoccurs, this approach

may

havelost sight

of

some

othercritical

dynamics

inthe

emergence

of

new

technologies.

Inparticular,

we

believe thattechnological developmentisnot theexclusive

domain

offirms

--orforthatmatter, acollectionoffirms ina givenindustry

—

butratheranactivity

which

cutsacross

many

typesof public

and

privateorganizations. Furthermore,it isproposed thattechnological

development

may

come

about throughtheconcerted effortsofa

community

of researchers

which

forms overtimeandthat spans thesediverseorganizations. If thisis indeedthe case, then the

progressof suchcommunities

may

be betterunderstood through a carefulexamination of

how

they

function. Thisfunctioningis

now

furtherexploredin thecasesofpolypropylene

and

epdm

technologies.

The

choice ofthesechemicaltechnologiesfulfillsthreegeneralresearch interests. Firstofall,

up

till

now,

the role of

R&D

communitiesintechnologicaldevelopment has beenstudiedfor a varietyofelectronics and

computer

technologies (e.g.,Josephsonjunctions, gallium arsenide

semiconductors, magnetic bubble memories,microgravity crystal growth, neural network

rubber technology are thefirstchemical technologies

on

ourresearch agenda. It is thus interesting

tosee

what

role

R&D

communities playin the fieldof chemicals. Second, polypropylene and

epdm

rubber technology havequite a differenthistory.

The

development

ofpolypropylene

technology has beencharacterized byquite a

number

of breakthroughssinceitsonset intheearly

1950's.

These

havehad atremendous influence onthe efficiency andthecost ofpolypropylene

manufacturingprocesses. Recent advancesincatalystsystems

and

process technology enable

reductionsinplantinvestmentof

more

than

50%

compared

toprevious practice.

Epdm

rubber technology,

on

theother hand, has been characterized

by most

expertsas a rather "stagnant

technology".

No

real breakthroughs have been

made

since itsinception, although incremental refinements have increasedthe performanceof

epdm

rubbertechnology as well. Thus,a

comparative studyof both technologies enables us tofocus

on

the similaritiesand differences

between

the

two

R&D

communitiesassociated with them. Third, both technologiesare by

now

wellinternalizedwithinanindustrialenvironment.This adds animportant

dynamic

dimension to

the

R&D

community

conceptproposedabove. Itallows usto shed

some

light

on

thequestionof

how

R&D

communitiesevolveas theirmentalcreations

become

commercializedwithin an

industrial setting.

Polypropylene and

epdm

rubbertechnology: anoverview

The

empiricaldataforthispaperare providedby thejournalandpatent literatureontwo

chemical technologies.

The

focusison the developmentofmanufacturing technologiesfor

polypropylene and

epdm

rubber.

As

aconsequence,

no

attentionispaid to subsequentusesofthe

products

which

result

from

polypropyleneor

epdm

rubber manufacturing processes.

The

production ofpolypropylene and

epdm

rubberrests

on

two

fundamental, though intertwined,

components: the catalystsystem usedtopolymerizethe

monomers

andthe processtechnology in

which

thispolymerizationtakes place. Both areintertwinedinthesensethatthe type ofcatalyst

whetherthepolymerizationreactiontakes placeina gasphaseorliquidphase) usingslightly

different catalystsystems arecompetingfor licenseesinanoligopolisticmanufacturermarketplace.

Inorderto investigatethe role

R&D

communitieshave played inthe

development

of both

technologies, abriefhistorical accountofpolypropylene

and

epdm

isnecessary.

Both

technologies

have theirrootsin scientificandtechnological developments

which

tookplace in the early 1950's

and

which

eventually ledtothe formation ofa

new

branchof organicchemistry anda

Nobel

Prize

foritsfounders, Ziegler

from

the

Max

Planck Institiit furKohlforshungin Mtilheim andNatta

from

the Politecnicodi Milan, in 1963.

The

earlyyears ofZiegler

-Nana

chemistry

The

origins ofthe polypropyleneand

epdm

rubbertechnologies

which

arepresently in use

go

back

to the

work

ofZieglerandNattainthe 1950's:

"The

discovery ofstereoregularpolymers

was

foreshadowed

by

certainevents in

polymer

science and

by

aborteddiscoveriesthat, like

minor

tremorspreceding a

majorearthquake,can be seen afterwardsas signalling andtriggeringearth-shaking

events. Karl Zieglerand GuilioNattaare justly

famous

fortheirepochaldiscoveries:

Zieglerforhis linear,crystallinepolyethylene, Nattaforhis isotactic, crystalline

polypropyleneandother stereoregularpolymers. YetZiegler

was

not the firstto

make

a linearpolyethylene, andstereoregularpolymers

were

postulated, prepared

and

publishedprior to Natta'swork.Linear, crystalline polyethylene was,in fact,

made

halfacentury before Zieglerbyanother

German,

named

von Pechmann."

(McMillan, 1979)

However,

Zieglerand Nattawereable to arise massiveindustrialinterest for theirdiscoveries.

Natta'scloselinkswith Montecatini (laterto

become

Montedison)

and

Ziegler's agreements and

relationships withsuch companies asHoechst

and

Hercules

Powder

Company

(aresult

from

the

Du

Pont

break-up followingananti-trust litigation)areoftremendous importance forthe structure

ofthe

R&D

communitiesas

we

willobserve

them

forboth technologies.

When

von

Pechmann

announced

his productatthe time, it

was

regarded a

mere

laboratory curiosity

and

itdid not receive

catalysts

and

Natta's subsequentpolymerization of propylene, however, generated research activity

and

plant construction allovertheworld.

Polyethylene

was

alreadya commercialreality and an

enormous

success

20

years before

Ziegler'sdiscovery. It was, however,a different kindof polyethylene, notlinear but "branched".

ICI

was

theleading licensor forthis high-pressure polyethylene process (Freeman, 1982) in

which

discovery both serendipityand

Ids

close linkswith the

Thermodynamics

Laboratory atthe

UniversityofLeiden played an importantrole (Reader, 1975). Ziegler'sdiscovery of

organo-metallic catalysts

made

theproduction of low-pressure polyethylene (and polypropylene) possible.

His

new

process did not needthe

enormous

amount

of pressure ofthe ICI process. Ziegler's

catalyst

was

acombination of titaniumchloride

and

aluminum

alkyl. But Ziegler

was

notthe only

researcher

working

in thisresearcharea.

Du

Pontmissedthe (bloody) battle for priority rights

on

organo-metallic catalysisonly withaone month'sdifferencein time.

As

aconsequence,

researchersat

Du

Pont gotreallyfrustrated becauseof

company

patent policies

which

had refrained

them from

publishingtheirdiscoveries:

"Du

Pont'schemists had aright to feeldisappointedifnot

somewhat

bitter,

because they had

made

the

same

discoverythatearned

someone

else

fame

and

fortune.

The

legal and commercialaspects of these inventions prevented the industrialscientistsat

Du

Pontandinthe other

companies

from

publishing their

results andreceivingcredit fortheir scientific achievements. Perhaps in this

particular case, immediatepublication by the

Du

Pontchemistsmight have given

the

company

abetter legal position becausenot allpublicity

would

have

gone

to

Ziegler

and

Natta,

who

shared a

Nobel

Prize in 1963. Inanyevent, the

Polychem

scientistsled by

Gresham,

Anderson, and

Robinson

had

made

outstanding

contributions of

Nobel

Prizecaliber, buttheextremelycompetitive natureof

polymer

R&D

prevented

them

or

Du

Pont

from

receivingwidespreadrecognition or

monetary

rewards forthis work." (Hounshell

and

Smith, 1988)

Ziegler

and

Natta's discoverycreated a

new

branchof chemistry and, althoughthe initial

production processesfor stereoregularpolyethylene and polypropylene caused

more

problems than

originally anticipated, industry allovertheworld

was

simply excitedaboutthe

new

prospectsthat

licensees. Patent battles, however, have been

numerous

beforepolypropylenerightswere given to

(and lateron,redrawn from) Montedison (McMillan, 1979; Hounshell

and

Smith, 1988). In the

early years,

two

other

companies

were developing alternativeprocesses forthe polymerization of

propylene.

The

StandardOil

Company

of Indianapursuedthe

development

of

molybdenum

catalysts,thoughitlackedthedetermination topush itthrough. PhillipsPetroleum developedand

commercializeda

chromium-based

polypropyleneprocess. Although not assuccessfulas the

Montecatinidevelopments,a

number

ofplants were builtusing this process.

Ziegler-Nattachemistry

opened

upentirely

new

vistas forboth polyethylene

and

polypropylene.

However,

synthetic rubber offered another outlet

The

searchforsyntheticrubber

has been going

on

for long(Whitbyet al., 1954). In 1860, Greville Williams succeededin

isolating a substance,

which

he

named

isoprene,

from

thedistillation productsofrubber. In

1933-34, the

German

IG

Farben

was

able topolymerizea

new

classofsynthetic rubbers,the Buna's.

Natta's findingthatcertainofhiscatalystscould be used to

make

random

ethylene-propylene

copolymersthat

were amorphous

and rubbery led the

way

to another type ofsynthetic rubbers, the

ethylene-propylenerubberclass.

Over

90%

ofthe ethylene-propylenerubbersare ofthe

epdm

type

which

means

that inadditionofthe basic ethylene-propylenecopolymer,they contain fractionsofa

third

monomer

which

usually isa diene.

Polypropylene technology, furtherdevelopments

Polypropylene has beenoneofthe success storiesofthe chemical industry. Exceptfor a sharp

recessionafterthe secondoilcrisisin 1974 anda suffering

from

an overcapacity in the beginning

ofthe80's, polypropylene has

known

adramatic growth in supply

and

demand.

Polypropyleneis

a bulkchemical. Itsapplications are diverse, forexample: batterycases, automobileparts, kitchen

show

thatdomestic

demand

forpolypropylene

grew

atan average annualrate of 1

1%

between

1971

and

1981, andat arateof

8%

in the last7 years(Chemical and Engineering

News, August

1989).

Of

the three

major

consuming

areas,growth has been fastest in

Western

Europe, followed

by

Japan and the UnitedStates.

Year

afteryear,

new

plants are

announced

to

meet

a growing

demand.

Two

basic groups ofproduction processescan be discerned (Kirk-Othmar, 1983; Encyclopedia of

Polymer

Science

and

Engineering, 1988): those

where

polymerization occurs inthe liquidphase

and those with a gas-phase polymerization.

When

lookingat apolypropylene plant,one findsthat

the polymerization

pan

isonlya fractionofthe totalinvestment.

The

downstream

investments are

over

50%

ofthetotalinvestment.

They

include: removal ofcatalystresidues, removalofatactic

polymer, andextrusionin ordertoreduce the solid

polymer

to

commercial

pellet-form. These huge

investmentscan be totally omittedifthe catalysthasthepropercharacteristics. Itisnot astonishing

then that catalyst

improvements

have

aimed

ateliminating those

downstream

processes.

Thissearch hasresultedin a

number

ofcatalyst "generations". Since the startof polypropylene

catalysis intheearly 1950's four significant breakthroughs incatalyst

development

can be

discerned together with

numerous

incremental

improvements

incatalystperformance

(Kirk-Othmar,

1983; Encyclopediaof

Polymer

Science andEngineering, 1988).

The

present superactive

third-generation catalysts (the

most

recentbreakthrough since thethird-generation catalysts) have

eliminatedthe purification,atacticremoval, and peptizationsteps

from

the manufacturingprocess

(Encyclopediaof

Polymer

Science andEngineering, 1988).

Processdevelopments have been going

on

eversince theinitialdiscoveries

by

Ziegler and

Natta. Today,essentially

two

process technologies are

competing

formarketshare

among

the 30

(which

was

initially

formed

as ajointventure between the Italian

Montedison

and

the

American

Hercules, although thelatterhasby

now

withdrawn

from

theventure;

Chemical

and Engineering

News, October

12, 1987

&

February, 1, 1988; Encyclopediaof

Polymer

Science

and

Engineering,

1988)

and

the Unipolprocess

by Union

Carbide

which

usesa catalystdeveloped by Shell

(European

Chemical News,

December

16, 1985; Burdett, 1986;Encyclopedia of

Polymer

Science

andEngineering, 1988). Theseare of course notthe only playersin an extremelycompetitive

environment.

However,

theirprocessesare with

no

doubtthe

most

important ones today. Other

important

knowledge

producers

on

polypropylene technology are the

German

firm

BASF,

andthe

Japanesefirms Mitsui Petrochemical Industries,Mitsubishi

Chemical

Industries

and

Mitsubishi

PetrochemicalIndustries.

The

present

Himont

success, forexample, has been preceded by a joint

catalystdevelopment between Montedison andMitsui Petrochemical (from 1975onwards).

Epdm

rubbertechnology,furtherdevelopments

From

atechnologicalpointofview,

epdm

technology hasevolvedless dramatically than

polypropylenetechnology.

No

significantbreakthroughs have occurred and in thepast 15 years

no

radically

new

processeshavebeen developed (Encyclopediaof

Polymer

Science

and

Engineering,

1988). In 1987, therewere 13 producers ofethylene-propylene elastomersin the world,excluding

Central Planned

Economy

countries. Thisadded up toa totalnameplatecapacityof1,347 million

pounds

peryear, (thiscan be

compared

withthe 1985 polypropylene

demand

of 17,380 million

pounds). Thus,

epdm

is

much

smaller than polypropylene asfaras

volume

production is

concerned, butitis therubber withthe fastestgrowthrate,

which

equals

6%

annually (Chemical

and Engineering

News, August

1989). Notwithstanding their beingthe fastest

growing

segment

among

the synthetic rubbers, ethylene-propylenerubbers arenotthe

most

important

form

of

synthetic rubber. Styrene-butadiene (theold Buna-S) andpolybutadiene rubbershit farbigger

At

the outset,researchersandindustrialiststhought

epdm

rubber

would

find ideal applications

in tires.

However,

thisdid not

work

out asexpected

and

has led toa vast

number

ofrather

fragmentedapplications nowadays, forexample: automotiveapplications such as radiator hoses

and

window

seals, wire andcable insulation,footwear, coatedfabrics

and

sheeting, belting and

rug underlays.

Epdm

rubberis

formed

through the copolymerizationof ethylene

and

propylene

togetherwithathird

monomer,

usually a diene.

The

catalysis uses organo-metallic catalystsofthe

Ziegler-Natta type.

The

resultofthisprocess, however, is a

random

and

amorphous polymer and

nota stereoregular

polymer

as inthecase of polypropylene.

Two

basic processesareusedfor

making

ethylene-propylene elastomers - solution

and

suspension.

The

solution process,

which was

thefirst

one

usedtoproduce commercialmaterial, is the

most

predominant. Recent developments

aim

atincreasingcatalystperformance. This isat leastpartlya consequence ofthe progress

which

hasbeen

made

with Ziegler-Nattacatalysts for the polymerization of propylene.

Major

epdm

producersare: in the

USA

-

Du

_{Pont, Exxon,}

_{Goodrich and}

Uniroyal; in

Europe

-

Bunawerke

Hiils (a jointventureofHills,

Hoechst

and Bayer),

DSM,

International Synthetic

Rubber

and Montedison; inAsia- _Japan

EP

_Rubber

_(which

was

_establishedasa jointventure between Japan

Synthetic

Rubber

and Mitsubishi Petrochemical), Mitsui Petrochemical and

Sumitomo

Chemical.

A

model

fortechnologicaldevelopment

Our

previous research intothedevelopment of

new

technologies,e.g. the

emergence

of neural

networktechnology, hasconvinced usthat thenucleus ofthis process isto alargeextentthe

work

of a dedicated

R&D

community

whichcreates a

momentum

forthe further

development

ofthe

new

technology. This

community

is

formed

andnurtured not

by

administrative decree, butbythe

autonomous

actionsofindividualresearchers

who

become

intriguedby an idea

and

are

committed

to solvingtheproblemsnecessary to

make

that ideawork. In asense, itisa self-organizing

process.

Out

ofthechaosofthe actionsofhundreds,

sometimes

even thousands, ofindividual

numerous

applicationsin severalbranches ofscience,

and

even inthescienceof science (Prigogine

andStengers, 1984). Itis our contention thatthe

same

principle can be applied tothe

emergence

of

a

new

technology. Thisself-organization impliesat the

same

timea subtle

form

ofcollective action. Researchersallovertheworldarecooperating

on

the

same

problem set, butatthe

same

time they

are

competing

forascarce

amount

ofrewards that willaccruetothose

who

arriveat

new

knowledge and

itsapplicationfirst. But even thoughresearchers

compete

for a limitedpool of

rewards, they

may

seethemselvesas

members

of alarger

community from which

theycan

draw

on,

and

contributeto,in a varietyofways. Mutatis mutandis,the

model

ofscientificevolution

depicted byHullcharacterizestechnology as well:

"Scienceisa matter of competitive cooperation, and bothcharacteristics are

important.

The most

importantsortof cooperation thatoccursin scienceis

the use oftheresults of otherscientists' research. This use is the

most

importantsortofcredit thatone scientistcan give another. Scientists

want

their

work

to be

acknowledged

asoriginal, butfor thatit

must

be acknowledged.

Their views

must

be accepted.For such acceptance, they needthe supportof

otherscientists.

One way

togain this supportis to

show

thatone's

own

work

rests solidly

on

precedingresearch.

The

desire for credit(i.e. competition) and

theneedfor support(i.e.cooperation) frequently

come

intoconflict.

One

cannot

gain support

from

a particular

work

unless

one

cites it, andthiscitation

automatically confersworth

on

the

work

cited

and

detracts

from

one's

own

originality." (Hull, 1988)(italicsadded)

We

propose thattechnological paradigms (Dosi, 1984)are selectedand gain

momentum

asa

result ofa similar interactionbetweencooperationandcompetition.

The

locusof paradigm

selectionis the

R&D

community. Scientistsandengineersseekcreditthroughthe developmentof technological

knowledge

and artefacts themselvesandthey give creditthrough theincorporation of

technological

knowledge

and artefactsdeveloped by other

community

members

into their

own

work.

To

use Kuhn's terminology (1970), the

members

ofthe

community

attempt to create a

shared "exemplar" as

outcome

oftheirselectionprocess.

The

discovery of organo-metallic

catalysts useful forlow-temperature polymerizationis a clear

example

ofsuch a creation.

The

new

11

stereochemistry. Ithas

spawned

researchactivity alloverthe world

aimed

ata betterunderstanding

of those particular catalystsand, at the

same

time,it hasaroused

enormous

industrialinterest.

However,

evenifthechemical world

showed

an immediateinterest onceZiegler

and

Nattahad

made

theirdiscoveries, ittook

some

time before researchers startedtoappreciate

what

they were

doing. Ziegler'scareeroffers a clear

example

ofthe dedication of an individual toa particular

researchcause andthe initial lackofappreciationofthatcause byother researchers in the chemicals

field.

A

subtle mixtureof competition, cooperation

and

controversyemerges.

When

Zieglerfirst

came

tothe

Max

Planck Institute forCoal Research he

made

itclearthat"he

would

not acceptanyoutside control in eitherchoice orpursuitof research goals" (McMillan,

1979). Zieglerhad never

done

anything with coal inhis lifeatthe

moment

thedirector'soffice

was

offered tohim, and he

was

clearthathe did not

want

to haveanything to

do

with coalfor therest

ofhislife either.

The

representativesofthe

Max

Planck Institute gave

him

cane

blanche to "play

around" with hisorgano-metallic

compounds.

Ziegler took an activestance in patenting,publishing

and

promoting his researchresults internationally.

However,

"As

sooften happens

when

one

scientist is askedtoevaluate another'swork,

Ziegler'shearers generallymissedthe

main

pointbecause they could seethe practicaldifficultiessoclearly.

At

one leading

US

industrial laboratory, for

example, the director solicited theopinionsofthose

who

had

heard Ziegler's

lecture, askingwhetherthere

was

anything worthfollowing

up

in

what

they

had heard.

The

uniformresponse

was

thatthe process

made

such amixture ofproducts thatthere

would

beaserious

problem

ofproductseparation

and

a

low

yieldofany specificdesiredchainlength. This disadvantage impressed

most

people

much

more

than theintriguing linearityoftheproducts.

Even

some

ofthe eminentrepresentativesofthe

German

chemical industry did notescape

theoccupational hazard oftheindustrial scientist: looking

upon

his

academic

colleagueas anintellectualdilettante

whose

ideas

and

work

areofdubious

practical import Bayer, oneofthe big

German

chemical companies, never took

alicense

from

Ziegler,although they had anearlyopportunity.

Pan

ofthereason

may

have beenthe attitudeexemplified by Bayer's research director.Dr. Otto Bayer.

He

was

among

a

number

ofleading

German

chemistspresentatadinner

which

Ziegler attended andat

which

he

was

asked aboutthe importance ofthe

new

chemistry

coming

outofthe

Max

PlanckInstitute. Ziegler, withcharacteristic dignified assurance,said he

was

sure thatit

would

stand asan important

contribution

and would

be

known

in the future as 'MiilheimerChemie'.

Bayer, polishinghisreputation for sarcastic wit,

remarked

thatthis

would

be anunfortunate choice ofa

name

foraninternationally

famous

process.

The

'Miill-eimer' (initial 'h' being silent inFrench). In

German,

Miilleimer

means

garbagecan." (McMillan, 1979)

Similardisinterestin Ziegler's favoriteresearch

was

demonstrated

by

the sponsors ofthe

Max

PlanckInstituteforCoal Research.

Upon

his request, they guaranteed

him

notonly

autonomy

in

directingresearch and freedomofpublication,butalso the rights to

any

usefulinventions thatfell

outside the fieldofinterestofthesponsoringcoalcompanies.

McMillan

adds: "His sponsors

would

have had

no

causetoregret thefirstofthese provisions hadit not been for the last.

They

obviously couldnot foresee the riversof

money

that

would

arise asaresultofZiegler's

work and

flowto

him

asaresultofthiscontract." (McMillan, 1979).

Of

course, noteveryone

showed

the

same

lackofinterestin Ziegler'sresearch program.

The

parallel effortspursued by

Du

Pontresearchers are aproof ofthis interest. Also, the

polymer

pioneer

and

authority

Herman

Mark

played the roleof a "prophet"

and

"publicist" as he

was

disseminatingZiegler's researchthroughout the

community

of

polymer

researchers.

The

earlier

yearsof Zieglercatalystswerethen clearly

marked by

a subtle mixture of cooperation and

competition, and even

by

some

sparks of controversy. Hull (1988) considersthis mixture of

cooperation andcompetition

which

oftengoes alongwith overt orsilentconflict,as necessary to

scientific progress.

The

same

argument undoubtedly holds fortechnologicalprogress.

Our

researchinterest in thispaper, however, focuses

on

how

the

R&D

communities

in

two

particular instancesofZiegler-Natta chemistry, polypropylene and

epdm

rubber, have evolved

since these epochaldiscoveriesinthe earlyfifties.

As

mentionedpreviously, thechemical industry

was

very excited andalmostimmediatelyinternalizedZiegler-Natta chemistry. This industry

was

at

thattime,perhaps

more

than anyotherindustry, heavily researchoriented.

And,

ofcourse, it still

istoday.

Meyer-Thurow

(1982),describing the evolutionofthe

R&D

activities ofthe

German

firm Bayer, speaks ofthe industrializationofinvention. Hounshell and Smith (1988) demonstratethe

13

importantroleof corporate research atthe

Du

Pont

company,

though atthe

same

time they give

numerous

descriptionsofthe tensions

which

arise

between

the researcher's desire for

autonomy on

the

one

hand,

and

corporate goals andobjectives

on

the other hand.

The

conflictis

between

a

self-organizing technologydevelopment andacorporation -driven one. Scholarshave arguedthat

technology is local, contrary toscience

which

is universal (Allen, 1977).

However,

indications

exist thatopenness, evenintechnologydevelopment,

may

bebeneficial:

"It is also unrealistic tosee thetransistoras theproduct ofthree

men,

or of

one

laboratory, orofPhysics, orevenofthe forties. Ratherits invention required

the contributionsofhundreds ofscientists,

working

in

many

different places,

in

many

different fieldsover

many

years." (Braun

and Macdonald,

1978)

Similar considerations have been

made

forthe chemical technology:

"A

powerful impulsetothe growth ofthe world marketalso

came

from

the

decision oftheUnited Statescourtin 1952 to

compel

ICI tolicenseseveral

other

US

chemical firms,in addition to

Du

Pont, the original licensees.

Although

bitterly contestedatthe time,

on

the groundsthatthecourthad

no

jurisdiction overICI, thedecision

may

wellhave been a blessingindisguise

even for ICI, inthatitalmostcertainly ledto a

more

rapidgrowth of

new

applications." (Freeman, 1982)

We

agree thattechnological developmentisa

complex

process. Inour model, however,

we

want

to focus

on

thefunctioningof

R&D

communities in relation totechnological progress.

We

assume

thattechnology is,inessence, a

body

of knowledge.

With

thisassumption

we

follow

Arrow

(1962), Layton (1974), and Constant(1980).

Even

though the ultimate goal

may

be to

produce something, thecurrency of

R&D

communities is not so

much

actual things as itisthe

ideas, ortheories, about

how

and

why

things

work

the

way

theydo. Therefore, technological

development

canbe understoodasanintellectual processthat evolvesovertime,

whereby

new

knowledge

iscreated andappliedinordertoconstruct a

new

product orprocess.

The

central actors

in thisprocessare the individual researchers

who

become

dedicatedto solvingthe problems,

and

it

isthey

who

setthe process in motion with their efforts to create and apply knowledge. In the

(2) theytransform information into knowledge,orin other words,they solveproblems, and(3)

they

communicate

informationand

knowledge

toeachother.

The

central propositionofour

model

is then

mat

the rateof progress ina technology'sdevelopmentis afunction of

how

quickly

problems

aresolved, which, inturn, depends

on

the

amount

ofinformationproduced, the

number

ofdiversesolutionsattempted,

and

the extentto whichinformation

and

knowledge

is

communicated

among

researchers.

We

expectthatthe

more

information availabletoaresearcher, the

more

likely he istoarrive ata useful solution. Moreover,

we

anticipate that the

more

diversity

inthe types ofsolutionsattempted, the

more

likely that criticalsolutions will be found. Lastly,

we

hypothesize that

communication

between researchersenhances the probabilityoffinding useful

solutions.

The

amount

of information and

knowledge

available toa researcher

depends

upon

how

much

hecan producehimself, or receivein theprocess of

communicating

with others.

The

communication

of informationand

knowledge

impliesthata researchercan alsogather information

and

knowledge

produced byanotherresearcher,anddisseminate toothersthat

which

he produces

orlearns. Forthe

most

partinformationis

communicated

informally by

means

ofinterpersonal

conversations,whereas

knowledge

is

communicated

in the

form

of

documented

claims, such as

with thesubmission ofpaperstorefereedjournalsor patentapplications.

Thisprocessof

communication

of information

among

researchersis influenced bythe

existenceof organizational boundaries betweenresearchers andtheir (andtheirorganization's)

economic

interests. Organizational boundaries areimportant for

two

reasons: first, they give riseto

information asymmetries

among

researchers because they

impede

the flow ofinformation and

increase the costof information gathering.

As

aresult, organizational boundaries can slow therate

of productionof

knowledge

within a

community

by reducingthe

amount

of informationavailable

15

the extentitincreases the diversityof

problem

solutionspursued

by

the

community

as awhole,

since researchersindifferentorganizationsare likely to havedifferentinformation sets. This

phenomenon

isillustrated bytheexcerpts

from

the

Braun

and

Macdonald

(1978)

and

Freeman

(1982) studiesabove.

The

communication impedance

effectoforganizational boundaries is

overcome

viaresearchers

who

act as technological gatekeepers

-

thatis,researchers

who

tend to

communicate

with others indifferentorganizations (Allen, 1977).

Moreover, itisreasonable to

assume

thatresearchers arerational, in the

economic

sense that

theyare motivated byself-interest: thatis, they are eagerto solve problems because there are

rewards forthose

who

do.

The

researcher'sobjective is to

maximize

the

amount

of

knowledge

he

produces

and

can layclaimto before otherresearchers, because these claimshave potential value.

A

researcherneednotproduceallofthe

knowledge

requiredtocommercializea technology, as

longas his

own

knowledge

claimsare secured.

Given

thiseconomicallyrational behavioron

behalfofresearchers, the

communication

of information across organizational boundaries likely

occurs as a

form

ofquidpro

quo

(von Hippel, 1988).

In

summary,

we

argue thatthespeedat

which

atechnology developswill

depend

on

how

rapidly a

community

forms andgrows,

and

the structural

and

behavioral characteristics the

community

adopts asitmatures.

Some

characteristics

promote

cooperation

between

researchers,

while othersfostercompetition. Incombination, bothelements areimportant contributors tothe

swiftconductoftheproblem-solving process: thecompetitionthatbreeds withthe growth and

diffusion ofthe

community,

escalating the

amount

of information produced

and

increasing the

diversity ofsolutions pursued; and thecooperation

which

flows

from

a

community

information

grapevine, enabling researcherstoexchange notes

on

the latestexperiments

and

topursue solutions

The

general

model

proposedin the previousparagraphs is

now

furtherelaborated through a

comparative studyofthe growthofthepolypropylene and

epdm

rubber

R&D

communities.

As

we

clearlydescribed, those technologies havedeveloped, sincethe 1950's, within a

research-intensive, industrial context.

The

juxtaposition ofthe

two

technologiesallows foracomparison of

the structure and conductwithin

two

R&D

communities that have generateddifferential technological progress,although

from

acommercialpointof view, the

two

can beconsidered

successful.

Methodology

The

data presented in this paperwereobtained bystudying theextensive

body

ofscientific,

technicalandpatentliterature generatedby the polypropyleneand

epdm

researchersthemselves in

thecourseoftheirwork. Since

communication

among

its

members

isa definedcharacteristic ofan

R&D

community,

studying thedocumented, or formal,

communication

among

researchersisa

convenient

means

forgaininginsight into thefunctioningofa

community.

This literatureprovides

uswith a richnessofbackgroundinformation aboutthe technology itself,

and

subsequently about

the structure and behavior within an

R&D

community.

Inordertosystematize the study ofthe polypropyleneand

epdm

communities,

two

electronic

relationaldatabaseswere analyzed.

The

polypropylenedatabase contained 1383 abstracts of journal

articles,conference proceedings andpatents since 1955.

The

epdm

databasecontained 613

comparableabstracts.

As

explainedpreviously, ourliterature search

was

specifically oriented

towards polypropylene and

epdm

rubbermanufacturing technologies,

and

not towards specific

applicationsofpolypropyleneor

epdm

rubber.

The

present analysisdoes notfocus

on

the

well-establishedbibliometric

methods

foranalyzingcitation frequencyandco-citation clusters(for

example,see Hjerppe, 1978; Narin,

Noma

and

Perry, 1987). Instead, the retrieved abstracts are

17

Indeed, eachabstractprovides awealth ofreliable, unobtrusive informationaboutresearch

activitieswithin the

community.

The

firstadvantage is its longitudinal character.

By

studying an

R&D

community

overtime, it

becomes

possibleto

show

the

dynamic

patterns

which

emerge.

The

second advantage laysin the information obtained

from

the abstracts.

Each

abstracttells us

who

the

researchersare and

how

their

numbers

vary

from

year-to-year.

The

abstracts also

show

what

topics theresearchers areworking

on

and

which

organizations

employ

them.

They

alsoreveal die

ties

which

develop overtime

among

the differentresearchers andorganizations by looking atthe

co-authorship ofthe papers orpatents and by

showing

themobility of researchers betweendifferent

organizations.

By

examining this informationduringa three-decade time-span, one isable to

visualizethe structuralchanges in the

R&D

community,

andalso, to

draw

inferences aboutthe

behaviorofthe researchers dedicatedtothe particulartechnology.

Comparing

the patterns for the

rapidly progressingtechnology (polypropylene) with thoseobtained forthe

more

stagnant

technology

(epdm

rubber) allows usto

make

certainpropositions aboutthe influenceofstructure

and conductwithin a

community upon

therateof technologicalprogress.

Notwithstandingitsapparentreliability andunobtrusiveness, thejournal

and

patentliterature

does have certain non-trivial limitations.First, there are

some

technical limitations

which

may

be

cumbersome,

though theycan be overcome.Patent databaseshave theunappealing characteristic

thatnotallpatentscontain inventornames. This is mainlya

problem

withJapanesepatents

and

with olderpatents.

We

were

able toresolvethis

problem

through thecombination ofseveral

databases:Derwent, Dialog and Inpadoc. This,however, can be a

cumbersome

process.

Second, patents

which

have been appliedfor, though nevergotissued,

do

not appearin

electronic databases.

However,

as

we

are only interested in detecting active

members

of

R&D

tocapture

major

trendsin the structureand conductof

R&D

communities. It is therefore not

necessary to identifyevery possibleparticipant.

A

third, andoften heard,objection tothe use ofthejournal or patentliterature states thatthe

sheer quantityofarticlesorpatentsdoes notnecessarily tell you whichorganizations are

most

active in acertainarea (Lambert, 1989).

However,

as

we

countinventors

and

not

numbers

of

articles orpatents, thisobjectiondoes notholdfor the

methodology

claimedhere.

A

fourth, and

somewhat

similar

problem

may

occurwith respecttoinventor andauthor

names

(Rappa, 1989).

Although

database services tryto standardizeauthorand inventor names,thisis

notalways the case. Thus,theresearcher himself has to

make

sure that

names

are standardized.

This

may

be atediousjob, though given our use ofthe literature data, itis avitalone.

Fifth, as faras patents are concerned,there are intercountrydifferences in thespeed ofissuing

patents andinthe timelagbetweenthe filingofthe patentanditspublication(although 18

months

is arather

common

delay periodfor

most

countries).

The

strategy

we

usedto alleviate this

problem

consistedintaking theprioritydate as a timeindicator,insteadoftheissue date.

The

priority date is

thusconsideredindicativeoftheyearin

which

theresearchers identified

on

thepatentwere active

inthe field.

Given

publication andpatent issuelags,our

methodology

also

makes

itimpossible to

obtainfull information

on

what

happened

during the last

two

years.

Once

past 1986, the data

do

nottellus

much. However,

we

assume

that thisis a

common

problem

facingeveryone

who

uses

thepatent literatureasa sourceofdata.

The

notion of

R&D

community

implies ablendingofscienceandtechnology, ofacademic

and

industrial pursuits,

and

furthermore, suggests asituation

where

papers andpatents are essentially

19

is anincreasingpopular

theme

in scholarly literature (Latour, 1987).

The

existenceofa

body

of

literature in technologydevelopmentisright atthe heartoftheconceptof an

R&D

community.

A

community

implies adegree openness,

which

findsanoutletin the

documented

literature.

Our

data

on

previoustechnologies

show

thatresearchers

do

indeed publish,regardless ofwhetherthey

residein an academicorinan industrial setting. Moreover, the

same

data

show

thatthe industrial

laboratoryiscertainly not the only locus oftechnology development. Universities

and government

laboratoriesplayprominentroles as well. Despitethe

argument

oftechnology being local, industry

scientists andengineers

do

indeedpublish, whethertheir

company

encourages

them

to

do

so or

not. This behavior has

many

advantages.

Companies

inrapidly

moving

fields often

do

research in

orderto stayupwith

them

andto havethecapability toexploitdevelopments in atimely manner.

Thereforethey havetojoin in the relevant

community,

andthis implies asharing of

knowledge

(Nelson, 1989).

The

documented

literatureisone possible

means

toachieve thisend. Moreover,

publishing researchresultsalso enhances

company

reputation

and

visibility,and it

may

atthe

same

timebeasubtle

form

ofadvertising. Thislastaspectisvery wellvisible inthe chemicals under

study.

The

majorlicensors(such asMontedison-Himont-Mitsui) tendtopublish a lot.

One

ofthe

reasonsforthisextensive

amount

publication (in refereedjournals!) surely is to praise the

versatility andqualityoftheirtechnologicalprowesses. If

company

researchers actively participate

in the

community,

and perhapsreceive

some

prestigious rewards such as

Nobel

Prizes, then their

firm alsobenefits

from

the

image

thus created.

Although

thispublication behaviorunderpinsour methodological approach touse the literature

as a source of datatoexplorethe characteristicsof

R&D

communities,italso points toanother

limitationofthe methodology.

The

formalcommunication,

documented

injournal

and

patent

literature,

may

bejustthe topofthe information-sharingiceberg.

Formal communication

iswithout

doubtrelatively limited

compared

tothe

amount

of informal

communication

that likelyoccurs

20

consequence, individuals participatingin the

community

may

not

become

visiblein the

documented

literature. Furthermore, there

may

also beactivity in developing the technology that is purposely

hidden

from

public

view

by those

who

seeit in theirbest interest to keepsecret. It isourfirm

belief, however,that this lasttypeofbehavioris not necessarilyconfined toindustry.

We

are

convinced(althoughitis still a hypothesisatthe

moment)

that academicscientists

and

engineers

may

alsohave personal incentives nottoreveal allthe information they possess.

Nevertheless, giventhese limitations, the literature can proveto be remarkablyuseful in

obtainingan initial understanding ofthe structural

and

behavioral

dynamics

of

R&D

communities.

The

analysispresented here seeksto limitthe obviousdeficiencies of usingthe

documented

literature by minimizingthe sensitivity topublicationfrequency, by not attempting toascribe

more

orless importancetoaparticularpublication orpatent,

and

byanalyzing thedatahistorically. Thus,

forexample,

we

arenotseekingto understand absolutemagnitudes so

much

as the

dynamic

trends

overtime,

and

we

are notseeking touncovertechnical secretsso

much

as obtaininga basic

understanding ofthepeople involved, the nature oftheirwork,

who

they

worked

with,

where

they

worked,

and

when

theyworked.

Taken

together, the dataobtained

from

theliteraturecan providea

comprehensivepictureofchange over time withinthe communities underinvestigation.

The

polypropylene and

epdm

communities, acaseofdifferential technology

development

Of

the 1383 records inthepolypropylenedatabase,

555 (40%)

arepublication abstracts while

828

(60%)

arepatent abstracts.

Of

the 613

epdm

records, 135

(22%)

are publication abstracts

while

478 (78%)

originate

from

the patentliterature.

The

fact thatthemajority ofthe retrieved

abstracts belongsto the

body

ofpatent literature

may

be anindication ofthe degree of

internalizationof both technologies within an industrialsetting.

At

the very beginning ofthispaper,

we

outlined thatthe

two

chemical technologies mightoffer interesting insights into

how

R&D

framework

ofa particular industry. In case of

newly

emerging

technologies, such asneural

network technology

(Rappa

and Debackere, 1989), the presenceofpatents within the

body

of

documented

literatureis negligeable oreven non-existent.

As

the technologydevelops and

becomes

internalizedwithin an industrial,competitiveenvironment, theopenness and

communication

throughthejournal literature

may

well shiftinfavorof

communication

through thepatent literature.

Although this

may

pointtoan increase in"secrecy" considerations,

we

believe thatitstill

remains an instrumentofholding together the

community.

Patents areintellectualclaims inthe

same

sense asarticlesareintellectual claims. Itisa

means

ofsharing knowledge, thoughit

may

not

havethe

same

public

good

characteras thejournalliterature (Sherer, 1980). Moreover, the

presence of journalabstracts inourdatabases

shows

that (as well inindustry as inacademia),

researcherskeeppublishingresults in the

open

literature,

and

atthe

same

time die useofpatents is

notconfinedtoindustry.

Academic

researchers takeout patents aswell atan increasingrate.

Finally,

we

must

notforgetthat

we

are dealing withchemical technologies.

The

abundantpresence

ofpatents

may

well be specifictodie chemicals field(Sherer, 1980). Nelsonreachesa similar

conclusion:

"While I wanttoemphasizethatpatentsplay a

much

smallerroleinenabling innovators toreapreturnsunder

modem

capitalism, there are certain industries

where

patent protection isimportant, perhapsessential, forinnovationincentive.

Our

questionnaire revealed

two

groups ofindustries ofthissort.

One

consistsofindustries

where

chemical compositionis a centralaspectofdesign: pharmaceuticals,industrialorganic chemicals, plasticmaterials, synthetic fibers, glass.

The

otherconsists ofindustries

producingproductsthatone mightcalldevices:airand gascompressors, scientific instruments, power-driven hand-tools, etc. In bothcases, thecomposition ofthe

productsisrelativelyeasy todefineandlimit." (Nelson, 1989)

Thus, die important shareofpatents inourdatabases has at least

two

differentreasons. First,

patents

may

have

become

a

more

appropriate

way

of sharingknowledge, giventhe internalization

22

conducive to the use ofpatents.

Anyway,

we

can safely

assume

thatpatents are just one possible

means

ofsharingknowledge. For ourpurposes, itcertainly

makes

no

difference whether

we

use

patentsor journalabstracts asa

means

toidentify the respective

R&D

communities.

The

differential

growth

of polypropylene

and

epdm

communities

An

analysisof both databases

was

conductedtodeterminethe

number

of individual researchers

worldwide,

who

were

active eachyearin the research

and development

ofthe technology. Ifan

individual is an author (orco-author) ofa journal paper, conference presentation or patentonthe

subject ofpolypropyleneor

epdm

technologyina given year, he isincluded as a

member

ofthe

community;

and hecontinuesto be includedasa

member

solongashe continuesto be an author

from

year-to-year. In this way,

membership

in the

community

is not sensitive to the

number

of

publicationsorpatents

by

an authorin a given year.

The

growthprofile of both

communities

is

shown

infigure 1.

The

datacoverthe timespan

from

1956 till 1986.

They

thus coverthe

development

ofboth technologies,once thebasic

paradigm

(Ziegler-Nattaorgano-metallic

catalysis)

was

firmlyestablished. Moreover, as

remarked

previously, the lackofpatent datadue to

publication lags

make

itimpossible tostretch the analysis

beyond

1986.

The

growth profile of bothcommunitiesis pretty

much

the

same

at the beginning, though the

polypropylene

community expanded

alotfasterthanthe

epdm

community.

The

epdm

community,

afteraninitial increase, remainedrather stagnantuntilthe beginning ofthe 1980's.

The

polypropylene

community

manifests acontinuous growth throughoutits history. Thisis also the

(0 « CO cc -Q E 3 250 n 200 150 -100 -50 -1956

PP

I ' ' ' 1981 1986

EPDM

PP

CORE

EPDM CORE

Figure 1:

The

growth ofpolypropylene and

epdm

communities

andassociated core groups, 1956-1986

However,

contrary to

what

we

haveseenin

newly

emergingtechnologies such asneural

networks

(Rappa and

Debackere, 1989), the growth is rather steady,with

no

realmanifestation of

a

bandwagon

effect.

A

closerexamination ofthe polypropylene literature (Kirk-Othmar, 1983), for

instance,reveals

two

importantcatalystdevelopmentsthathave startedinthe 1960's: theuse of

electrondonors and theuseofa

magnesium

support for the titanium chloride catalyst.

However,

they

became

only included in thepolypropylene maufacturing processesin the late 1970's, i.e.

with thedevelopment ofthe third-generation catalysts andits subsequentsuperactive catalysts

(Encyclopediaof

Polymer

Scienceand Engineering, 1988).

We

could findan article

on

electron

donors as early as 1965, whileafirst

development on

the use ofcatalystsupportsystems could be

24

polypropylene.

On

average, atime lagof 10 years occurred beforetheinitial catalystdevelopments

were

vigorously incorporatedinto industrialprograms.

However,

each development,

accompanied

by amultitudeof small incremental improvements,

may

have arousedinterest in polypropylene

catalystresearch

and

may

thus have causedother researchersto start

working on

the subject.

The

advent ofthe superactivethird-generationcatalystsin the secondhalfofthe 1970's hascaused a

similar increasein interest. Thus,certaindevelopments inthe stateof technological

knowledge

may

actas atrigger forresearchers toenter the

community. These

considerations bring usto

two

central

questions

on

ourresearch agenda.

The

firstcrucialquestion,oftenraised

by

policy

makers

when

looking atsimilar data, is oneof

causality:

Does

technological progressoccurasaconsequenceofanincreased

manpower

effort

on

behalfofthe

community

who

ultimatelyproducestheadvances, oristhe increasein the

number

of

researchers aconsequenceof technologicaladvances

which by

and of themselves generatea

brighteroutlook forthe technology underscrutiny?

We

believetherealityis amixture ofthe two.

Our

data suggestthattriggers are necessaryto arise at least

some

interest.

The

discoveriesby

Zieglerand Natta

were

one sucha trigger.

The

suggestion thatelectrondonors or

magnesium

supportmight prove valuablecan be considered another trigger.

However,

the

most

these triggers

usually offerisa

box

of

Pandora

fullof"problems still to besolved".

At

that

moment,

we

believe

anincreasein

manpower

within die

community

really

becomes

a necessity

and

a determinant ofthe

subsequentrateoftechnological progress.Theissue isthenreduced to theissueofwhetherthe

problems inPandora's

box

can catchthe imagination

and

elicittheinterest of individual

researchers. Ifitcan, the future ofthetechnology will look

much

brighter.

The

contrastbetween polypropylene and

epdm

seems

enlightening in thiscontext.

Polypropylene technology andcatalysis haveclearlybeen ableto lure researchersinto their linesof

inquiry. Breakthroughsandincremental

improvements

have accumulated eversince.

Epdm

started

seem

appealing

anymore

andinterest in thetechnology stagnated. This

remained

largely so

throughoutthe 1960's

and

1970's. "Stagnanttechnology"

was

the connotationsubsequently

reserved for

epdm. However,

recentiy,a cross-fertilization between bothpolypropylene and

epdm

communities

seems

tooccur.

The

progress inpolypropylene technology,mainlyin the

domain

of

catalysts,

now

provides an impetus totake up

on

epdm

catalysts

more

vigorously research.

The

polypropylenedevelopments have

somehow

created the beliefthat, since

epdm

usesan analogous

typeoforgano-metalliccatalysts, the

example

ofpolypropylene can berepeatedin

epdm.

The

coming

years will

show

whetherdie increase insize ofthe

epdm

community

will persistandwill

be able tofulfillthe expectations. Ifnot,

we

may

expectonce again adecline ofthe size ofthat

community

and

a hostofresearchers switching toother(research) pastures.

The

secondimportant question revolvesaroundthedegree to

which

theevaluation ofthe

contentofPandora's

box

andthe subsequentdecision

on

whetherthiscontentisinteresting

enough

toenterthefield of, for

example

catalystresearch, isan

autonomous

decisionofindividual

researchersorto

what

degreeit is

management

driven.

Our

previous studies

on

newly emerging

technologieshintinthe direction ofaquasi-complete

autonomy

onbehalfofthe individual

researchers. In instances as neuralnetworktechnology, the

community

then

becomes

the primary

locusof technological progress (Rappa and Debackere, 1989).

The

direction the technologytakes

andthe problems

which

are consideredworthwhile are both heavily influenced

by

consensus

and

controversiesthatreignwithin the broader

R&D

community.

Our

data for the

two

chemical

technologies suggest aslightly, thoughnot completely,different patternonce the technology

becomes

internalizedwithin anindustrial environment suchas diechemical industry.

The

internalizationofboth technologies withinan industrial setting,andthe factthatdie initial

development

ofsignificant catalystimprovements alloccurredwithin industriallaboratories (such

as Shell,Montedison,Mitsui) limit toa certain degreethe

freedom

of individual researchersto

26

the availabilityofapilot plant), theinterests ofthe individualresearchers are to

some

extent

dependent

upon

theinvestment decisions ofthe largechemicalcorporations that

form

theprimary

locus for catalyst research.

However,

even ifthe industrial world cancontrol theentry rateofresearchersinto

polypropyleneresearch, a certain level of

autonomy

remains. Firstofall, ourdata

show

that

we

have been able toidentify a group ofresearchers

who

share

knowledge

(through the

documented

literature) about theirresearchefforts. In themethodological partofourpaper,

we

discussedthe

importanceofthis

knowledge

sharing as aprimaryindicatorof

community

existence. Second.

even iflargecorporations are takingdecisions

on

enteringorquittinga certain line ofcatalyst

research, they cannot

do

this withoutthe input

from

individual researchers. Theiropinion,

however,will not be solelyintermsofcorporateobjectives.

They

are researchers, having a stake

atthe

community

level themselves. Thus,

pan

oftheirdecision will almost certainly be based

on

what

they perceivethewider

community

isjudging aworthwhile researchavenue. Third,

when

lookingat thedatain figures2and 3,

we

seethatthe industrial world onlyrepresents

pan

ofthe

community.

The

other

pan

consistsof public sector organizations. This group isa mixture ofuniversity

and

government

laboratories.

Here

corporate objectives are non-existent.

The

researchers at

academic

laboratoriesnormally have great

autonomy

indeciding

on

the lines ofresearch they want

topursue.

We

then seethatpolypropylene has been able toarouse alotofinterest within

non-industrial settings.

Cenain

triggers have been abletolureacademlicresearchers into thefield. This

has, for instance, beenthecase withthe

emergence

ofsuperactive third-generation polypropylene

catalystsin the late 1970's. This presence ofacademicresearchers has ledto subsequent

developments

inpolypropylene catalyst research.

A

good example

is

Kaminsky's

(University of

remarkablydifferent.Here, public sectorinteresthas generallybeen low.

The

research

community

was

able to

improve

on

the technology, but

compared

topolypropylene these

improvements

resemble

more

a status-quo. Publicsectorresearchersobviously did notuse their

autonomy

toopt

forthisline ofresearch.

250 CO 0> CO 4> 0C O E 3

Figure 2:Sectoral distribution ofpolypropylene

researcher

community, 1961-1986

To

conclude,

we

can say thateven incaseofthetechnologies

examined

in thispaper(and

that are closely tiedtocorporateinterests),

we

areableto identifyaworldwide group ofpeople

who

at leasthavethe potential toinfluence the direction technological developmenttakes.

Moreover, there certainlyexistsacorreletionbetween theprogress of technologicaldevelopment

andthe growthofthecommunity.Thisis exemplifiedby thecomparisonofpolypropylene and

epdm

developments.It is, however,impossibleand presumably

wrong

to suggestthat thereis a

28

the

form

ofdevelopments

which

offer

enough

challengingproblems forresearchers to take the

decision toenteracertainlineofinquiry.

The

greaterthe

number

of researchers (

and

organizations

)

thatcanbe attracted to a particular research agenda,the greater the likelihood thatproblems will be

solved

more

quickly orthataconsensuswillbe reachedthattheproblems cannot be solvedatthe

moment.

M 3

o

125 100 -Public Industry

Figure 3:Sectoral distributionof

epdm

researcher

community,

1961-1986

Cycles of enthusiasm

and

despair

The

growth ofthepolypropyleneresearch

community

israther steady, progressing

from

its

onsettill the 1980's(figure 1). Since the

community was

able to sustain the

momentum

of

technologicalprogress, there existednorealreasons forresearchers to getdespaired aboutthe